Skip to content

Prompts

Note: The links to the various NeonSkill methods in this documentation point to the mycroft-core documentation. There is no material difference between them for the methods referenced and, since there is not publicly available neon_core documentation, we will reference mycroft-core instead.

Here we look at how to implement the most common types of prompts. For more information on conversation design see the Voice User Interface Design Guidelines.

Get Response

Any Skill can request a response from the user - making a statement or asking a question before the microphone is activated to record the User's response.

The base implementation of this is the get_response() method.

To see it in action, let's create a simple Skill that asks the User what their favorite flavor of ice cream is.

from neon_utils.skills.neon_skill import NeonSkill
from ovos_workshop.decorators import intent_handler


class IceCreamSkill(NeonSkill):
    @intent_handler('set.favorite.intent')
    def handle_set_favorite(self):
        favorite_flavor = self.get_response('what.is.your.favorite.flavor')
        self.speak_dialog('confirm.favorite.flavor', {'flavor': favorite_flavor})


def create_skill():
    return IceCreamSkill()

In this Skill we have used get_response() and passed it the name of our dialog file 'what.is.your.favorite.flavor.dialog'. This is the simplest form of this method. It will speak dialog from the given file, then activate the microphone for 3-10 seconds allowing the User to respond. The transcript of their response will then be assigned to our variable favorite_flavor. To confirm that we have heard the User correctly we then speak a confirmation dialog passing the value of favorite_flavor to be spoken as part of that dialog.

Optional Arguments

The get_response() method also takes the following optional arguments:

  • data (dict) - used to populate the dialog file, just like speak_dialog()
  • validator (function) - returns a boolean to define whether the response meets some criteria for success
  • on_fail (function) - returns a string that will be spoken if the validator returns False
  • num_retries (int) - number of times the system should repeat the question to get a successful result

See it in action:

Video Tutorial: Asking a question

Yes / No Questions

ask_yesno() checks if the response contains "yes" or "no" like phrases.

The vocab for this check is sourced from the Skills yes.voc and no.voc files (if they exist), as well as neon-core's defaults (contained within neon-core/res/text/en-us/yes.voc). A longer phrase containing the required vocab is considered successful eg both "yes" and "yeah that would be great thanks" would be considered a successful "yes".

If "yes" or "no" responses are detected, then the method will return the string "yes" or "no". If the response does not contain "yes" or "no" vocabulary then the entire utterance will be returned. If no speech was detected indicating the User did not respond, then the method will return None.

Let's add a new intent to our IceCreamSkill to see how this works.

from neon_utils.skills.neon_skill import NeonSkill
from ovos_workshop.decorators import intent_handler


class IceCreamSkill(NeonSkill):
    @intent_handler('do.you.like.intent')
    def handle_do_you_like(self):
        likes_ice_cream = self.ask_yesno('do.you.like.ice.cream')
        if likes_ice_cream == 'yes':
            self.speak_dialog('does.like')
        elif likes_ice_cream == 'no':
            self.speak_dialog('does.not.like')
        else:
            self.speak_dialog('could.not.understand')


def create_skill():
    return IceCreamSkill()

In this example we have asked the User if they like ice cream. We then speak different dialogs whether they respond yes or no. We also speak some error dialog if neither yes nor no are returned.

See it in action:

Video Tutorial: Asking a yes/no question

Providing a list of options

ask_selection() provides a list of options to the User for them to select from. The User can respond with either the name of one of these options or select with a numbered ordinal eg "the third".

This method automatically manages fuzzy matching the users response against the list of options provided.

Let's jump back into our IceCreamSkill to give the User a list of options to choose from.

from neon_utils.skills.neon_skill import NeonSkill
from ovos_workshop.decorators import intent_handler


class IceCreamSkill(NeonSkill):
    def __init__(self):
        NeonSkill.__init__(self)
        self.flavors = ['vanilla', 'chocolate', 'mint']

    @intent_handler('request.icecream.intent')
    def handle_request_icecream(self):
        self.speak_dialog('welcome')
        selection = self.ask_selection(self.flavors, 'what.flavor')
        self.speak_dialog('coming.right.up', {'flavor': selection})


def create_skill():
    return IceCreamSkill()

In this example we first speak some welcome.dialog. The list of flavors is then spoken, followed by the what.flavor.dialog. Finally we confirm the Users selection by speaking coming.right.up.dialog

Optional arguments

There are two optional arguments for this method.

min_conf (float) defines the minimum confidence level for fuzzy matching the Users response against the list of options. numeric (bool) if set to True will speak the options as a numbered list eg "One, vanilla. Two, chocolate. Or three, mint"

See it in action:

Returning responses to the intent parser

So far we have looked at ways to prompt the User, and return their response directly to our Skill. It is also possible to speak some dialog, and activate the listener, directing the response back to the standard intent parsing engine. We may do this to let the user trigger another Skill, or because we want to make use of our own intents to handle the response.

To do this, we use the expect_response parameter of the speak_dialog() method.

from neon_utils.skills.neon_skill import NeonSkill
from ovos_workshop.decorators import intent_handler


class IceCreamSkill(NeonSkill):
    def __init__(self):
        NeonSkill.__init__(self)
        self.flavors = ['vanilla', 'chocolate', 'mint']

    @intent_handler('request.icecream.intent')
    def handle_request_icecream(self):
        self.speak_dialog('welcome')
        selection = self.ask_selection(self.flavors, 'what.flavor')
        self.speak_dialog('coming.right.up', {'flavor': selection})
        self.speak_dialog('now.what', expect_response=True)


def create_skill():
    return IceCreamSkill()

Here we have added a new dialog after confirming the Users selection. We may use it to tell the User other things they can do with their Neon device while they enjoy their delicious ice cream.